Land Use Mapping and Change Detection in the Coastal Zone of Northwest Mexico Using Remote Sensing Techniques
Keywords:
Multispectral images, remote sensing, supervised classification, Extraction and Classification of Homogenous Object algorithm (ECHO)Abstract
A multitemporal post-classification study with data from the Landsat Multispectral Scanner (MSS) and Thematic Mapper (TM) was made to detect changes in the landscape of the Majahual coastal system, along the Mexican Pacific. Six land-use classes were used as direct indicators of the landscape condition. Mangrove, lagoon, saltmarsh, dry forest, secondary succession, and agriculture were the categories selected to evaluate the changes by comparing four thematic maps (from 1973 to 19971. The accuracy of the classification (only in the 1997 scene) was calculated from an error matrix, using the overall accuracy assessment (70%) and the Kappa coefficient (0.61). Both values indicate that the agreement in the classification was moderate, but better than one obtained by chance. The analytical comparison of data sets (1973 vs. 1986, 1986 vs. 1990, and 1990 vs. 1997) was done by using a change detection matrix and the Kappa coefficient. Agreement between data sets varied from 61% to 68%, all moderate but enough to determine the general trends of change in the system. These are mainly typified as loss of natural cover (especially dry forest) and the fragmentation of the landscape, with agricultural activities and their subsequent effects (secondary succession, modification of drainage patterns) the main transforming agents.